Dear Listmembers, we´re running an Europa 20-20 MS in CF-mode for analyses of O-isotopes. The MS itself produces a noise signal in the range of 0.05-0.1 permil (18O). However, I always observe a drift over 3-4 hours in the range of 1-2 permil. The drift is pretty linear an can be well fitted. The 1sigma error of drift corrected analyse of 0.3 mg silicate is between 0.12 permil. Does anybody works with such a machine and observes this drift? Another problem is the poor (unfortiunalety MS-Windows-based) Europa software ("vanca-sl 1.99" and especially "ancrepro" evaluation software). If there is anybody having problems with calculation the results using the ancrepro-programm, he may send a mail to me. I´ve written a small programm (in C, which I can compile for MS-Windows and Linux) for the evaluation of the raw data. It produces an *csv-file which can be read in Excel or StarOffice, Andreas Dipl. Min. Andreas Pack Mineralogisch-Petrologisches Institut und Museum Poppelsdorfer Schloss D-53115 Bonn (Germany)